In tennis sports, improving athletes’ competitive performance is the fundamental purpose of sports training. Physical fitness, as an important part of athletes’ competitive ability, is the most basic and development potential training factor that athletes must have. In order to accurately assess the association between tennis players’ physical training and their athletic performance, this paper uses the gray correlation analysis method, which is combined with the TOPSIS method to form the TOPSIS-gray correlation analysis model in this paper. In this paper, 170 athletes from a provincial tennis training camp were selected as research samples, and the combined model was used to explore the potential correlation between physical fitness structure and tennis athletic performance, and analyze the more important nine physical fitness indexes that should be achieved by tennis athletes. Thus, a set of physical structure characteristics and hierarchical model for tennis athletes was constructed, and the data analysis of the model showed that the top three selection criteria for tennis athletes, in descending order, were 150m running performance, 60m standing start performance and arm length. Combined with the model analysis data, this paper develops a set of physical structure model scoring table for tennis players. The 150m running performance should be no more than 16.73 seconds, the 60m standing start performance should be no more than 8.65 seconds, and the arm length should be no shorter than 75.68 centimeters. The TOPSIS-gray correlation analysis model and model rating scale established in this paper can make individual gray correlation analysis of tennis sports players, rate them and compare the overall highest level, so as to find the specific short board of each athlete. To provide scientific theoretical reference for their later physical training, in order to improve the athletes’ competitive level and performance.